Here is a report run by hand on Win98, in which I avoid running any tests. Automated smoke report for 5.9.0 patch 22203 on MSWin32 - 4.10 (x86) (7106) using gcc version 3.2.3 (mingw special 20030504-1) Report by Test::Smoke v1.18.10 (perl 5.8.2) [19 minutes 53 seconds]
O = OK F = Failure(s), extended report at the bottom X = test(s) failed under TEST but not under harness ? = still running or test results not (yet) available Build failures during: - = unknown or N/A c = Configure, m = make, M = make (after miniperl), t = make test-prep 22203 Configuration (common) -DCCHOME=C:\m -DCCTYPE=GCC -Dgcc_v3_2 ------------ ---------------------------------------------------------------- ? ? ? ? | | | +----- PERLIO = perlio -DDEBUGGING | | +------- PERLIO = stdio -DDEBUGGING | +--------- PERLIO = perlio +----------- PERLIO = stdio Summary: PASS-so-far MANIFEST did not declare 'Policy.sh' MANIFEST did not declare 'win32/.makefile.mk.swp' MANIFEST did not declare 'win32/.smoke.mk.swp' MANIFEST did not declare 'win32/makefile.mk~' MANIFEST did not declare 'win32/smoke.mk' MANIFEST did not declare 'win32/smoke.mk.old' MANIFEST did not declare 'win32/smoke.mk.tmp' I did some cheating. I rewrote smoke.mk with group recipes, rather than create makefile.95. I guess I should now try sticking closer to the usual install procedure and add IS_WIN95 options in Test::Smoke::Util. I also monkied around with makefile.mk. Patch 22203 doesn't contain perlexe.ico. As for the tests, in earlier smoking attempts I haven't been able to run all the tests. I get further with a Bourne shell than with command.com, but even running dmake test from the command line in the mingw shell, although I get a test report, dmake hangs after that and doesn't exit. Running from the command line in command.com, dmake hangs quite early in the tests at comp/cpp. There are various other one-line things I did to get this far, but nothing worthy of a patch I think. I am interested in getting the tests to run, or at least removing the tests that hang the smoke, but need an alternative to trial and error methods to do it, because of the number of the tests. -- Greg Matheson, Taiwan
